A data acquisition and analysis method and system for safety valve calibration

Abstract

The present application belongs to the technical field of data acquisition and analysis, and specifically relates to a data acquisition and analysis method and system for safety valve calibration, comprising: obtaining multi-dimensional original signals of safety valve calibration, imposing condition constraints based on a time synchronization mechanism on the multi-dimensional original signals, performing deep analysis and feature mining on the multi-dimensional original signals, extracting characteristic parameters reflecting dynamic response characteristics of the safety valve, establishing an individualized performance baseline file based on historical data of previous calibrations of a single safety valve, performing correlation analysis on the extracted characteristic parameters, a factory initial parameter and a historical calibration record, and constructing a performance evolution track reflecting the state of the single safety valve from factory to the present; the present application can realize complete capture and reservation of calibration process information, dynamic matching of calibration excitation and actual working conditions, accurate tracing and distinguishing of calibration abnormalities, and conversion from post-event eligibility determination to pre-event predictive maintenance.

Description

Technical Field

[0001] This invention belongs to the field of data acquisition and analysis technology, specifically a data acquisition and analysis method and system for safety valve calibration. Background Technology

[0002] The existing offline calibration of safety valves generally adopts the test paradigm of static uniform pressure increase on the calibration bench. By gradually increasing the inlet pressure before the valve, core parameters such as the opening pressure and reseating pressure of the safety valve are collected. The measured values ​​are compared with the standard thresholds specified in the safety technical regulations, and the pass / fail judgment result is output. The process data such as the pressure time domain curve, valve disc displacement trajectory, and vibration response signal collected simultaneously during the calibration process are usually only used for single parameter calculations and are not archived or used for in-depth analysis.

[0003] However, the excitation method of a single uniform pressure increase is fundamentally mismatched with the dynamic loads such as periodic pressure fluctuations and start-stop impact pulses that safety valves actually experience during service. This results in the verification results failing to accurately reflect the dynamic response characteristics and operational reliability of the safety valve under actual operating conditions. Furthermore, a wealth of process data containing information on valve disc movement rate, opening and closing oscillation characteristics, and sealing surface contact status is not effectively utilized, making it difficult to establish a performance baseline for the entire lifecycle of a single safety valve to identify individual differences and gradual performance degradation. Additionally, there is a lack of effective mechanisms to distinguish between verification system noise and safety valve performance changes, making it impossible to accurately trace the cause of problems when a single verification result is abnormal. Therefore, this invention provides a data acquisition and analysis method and system for safety valve verification. Summary of the Invention

[0004] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0005] The technical solution adopted by this invention to solve its technical problem is: a data acquisition and analysis method for safety valve calibration, comprising the following steps: Step 1: Obtain the multidimensional raw signals for safety valve calibration; Step 2: Apply time synchronization-based constraints to the multidimensional original signal; Step 3: Perform in-depth analysis and feature mining on the multidimensional raw signals to extract feature parameters that reflect the dynamic response characteristics of the safety valve; Step 4: Based on the historical data of each safety valve's calibration, establish an individualized performance baseline profile, and perform correlation analysis between the extracted feature parameters and the initial factory parameters and historical calibration records to construct a performance evolution trajectory that reflects the safety valve from its factory state to its current state. Step 5: Based on the full lifecycle performance baseline file, perform long-term trend prediction and degradation assessment, calculate the comprehensive degradation index, and trigger an early warning when the rate of change of the comprehensive degradation index exceeds the preset threshold; Step 6: Calculate the remaining useful life based on the degradation trend extrapolation model and generate the judgment result.

[0006] Preferably, in step three, the multidimensional original signal is subjected to in-depth analysis and feature mining. The peak value of opening pressure, the value of reseating pressure, the pressure establishment time, the valve opening displacement, the action response delay time, the vibration main frequency and the frequency band energy distribution parameters are extracted through joint analysis in the time and frequency domains and wavelet multi-scale decomposition.

[0007] Preferably, in step five, the characteristic parameter sequence is fitted with a time series analysis algorithm to extract the rate of change and acceleration index of the performance parameters throughout the entire life cycle, identify the decay trend of valve disc movement rate, the increase trend of action response delay and the decline trend of sealing performance, and calculate the comprehensive degradation index using a multi-parameter fusion evaluation mechanism. A performance warning is triggered when the rate of change of the comprehensive degradation index exceeds a preset threshold.

[0008] A data acquisition and analysis system for safety valve calibration includes: The dynamic multidimensional data acquisition module acquires the multidimensional raw signals of the safety valve verification and applies conditional constraints based on a time synchronization mechanism to the multidimensional raw signals. The unsteady-state pressure excitation application module performs in-depth analysis and feature mining of multidimensional original signals based on joint time-frequency domain analysis and wavelet multi-scale decomposition, and extracts feature parameters that reflect the dynamic response characteristics of the safety valve. The multi-dimensional feature extraction module establishes an individualized performance baseline profile based on historical data from each calibration of a single safety valve, and performs correlation analysis between feature parameters and initial factory parameters and historical calibration records to construct a performance evolution trajectory. The full lifecycle baseline management module performs multi-parameter fusion evaluation on the performance evolution trajectory, calculates the comprehensive degradation index, triggers an early warning when the rate of change of the comprehensive degradation index exceeds a preset threshold, calculates the remaining service life based on the degradation trend extrapolation model, and outputs a judgment result that is bound to the qualitative status of the industrial site.

[0009] Preferably, the process of acquiring the multidimensional raw signal for safety valve verification further includes: Pressure acquisition module, which acquires the instantaneous value of the inlet pressure before the valve; The displacement acquisition module tracks the displacement trajectory of the valve disc during the opening and closing process; The vibration acquisition module captures the vibration response characteristics of the shell, and all acquisition channels ensure the time consistency of multi-channel data through a hardware-triggered synchronization mechanism.

[0010] Preferably, in the process of performing in-depth analysis and feature mining on the multidimensional original signal, the peak value of the opening pressure, the value of the reseating pressure, the pressure establishment time, the valve opening displacement, the action response delay time, the vibration main frequency and the frequency band energy distribution parameters are extracted through joint analysis in the time and frequency domains and wavelet multi-scale decomposition.

[0011] Preferably, in the process of establishing an individualized performance baseline profile, the multidimensional feature parameters extracted from each verification are correlated with the initial factory parameters, cumulative operating time, and historical verification records of the safety valve; Construct an anti-performance evolution trajectory and identify the initial differences between individual products of the same model by establishing a mapping relationship between characteristic parameters and service conditions.

[0012] Preferably, it also includes a noise separation and identification module, which, after acquiring the multidimensional original signal of the safety valve verification, performs independent component analysis on the mixed signal, separates the mixed signal into the equipment system noise component and the actual response component of the safety valve, and then establishes a noise benchmark reference model. Based on a noise spectrum feature library, the interference signals generated by the device itself are matched, identified, quantified, and eliminated. When a feature parameter is detected to deviate from its historical fluctuation range, an anomaly type determination conclusion is output.

[0013] Preferably, it also includes a degradation trend analysis module, which uses multiple time series models for parallel fitting in the process of multi-parameter fusion evaluation of performance evolution trajectory; The optimal model is automatically selected based on the fitting residuals, and the degradation rate and acceleration of each feature parameter are quantified. The comprehensive degradation index is calculated based on a multi-parameter weighted fusion algorithm, and a trend analysis report is generated.

[0014] Preferably, it also includes a prediction and early warning module. In the process of calculating the remaining useful life based on the degradation trend extrapolation model, the degradation trend extrapolation model adopts a combination architecture of gray prediction model and neural network prediction model to calculate the distribution of the remaining useful life. The prediction result includes point estimates and confidence intervals.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention, through a dynamic multi-dimensional verification data synchronous acquisition step and a full process data archiving mechanism, can achieve complete capture and retention of verification process information. It can change the limitation of traditional verification modes that only extract a single opening pressure value, so that the original data containing rich state information such as pressure time domain curves, valve disc displacement trajectories, and vibration response signals are no longer discarded. This provides a complete data foundation for safety valve performance evaluation and can effectively solve the problem of single evaluation dimensions caused by the lack of process data in existing verification technologies.

[0016] 2. This invention achieves dynamic matching between the verification excitation and the actual working conditions through the non-steady-state pressure excitation application step. It generates a composite excitation waveform containing periodic fluctuations, random pulses, and variable slope pressure rise to replace the traditional static uniform pressure rise excitation. This ensures that the verification load shape is consistent with the dynamic load that the safety valve bears in the actual high-pressure industrial system, thus solving the problem of the verification results being out of sync with actual reliability caused by the excitation shape mismatch in the existing verification method.

[0017] 3. This invention enables accurate tracing of verification anomalies through a noise and performance change separation and identification step. It employs an independent component analysis algorithm to separate the mixed signal into equipment system noise and the true response components of the safety valve. By using a noise benchmark reference model to eliminate equipment interference, it solves the problem in existing technologies where the cause of anomalies cannot be determined due to the lack of a mechanism to distinguish between noise and performance changes, thus ensuring the accuracy and reliability of the verification conclusions.

[0018] 4. This invention enables a paradigm shift from post-event judgment to pre-event prevention through the construction of a full life-cycle performance baseline and the analysis of gradual degradation trends. It establishes an individualized performance baseline file for each safety valve to track its full life-cycle evolution trajectory, and uses time series analysis algorithms to extract performance degradation trends and predict remaining service life, thus solving the problem that existing verification methods can only achieve post-event compliance judgment without predictive and early warning capabilities. Attached Figure Description

[0019] The invention will now be further described with reference to the accompanying drawings.

[0020] Figure 1 This is a structural block diagram of a data acquisition and analysis system for safety valve calibration according to the present invention; Figure 2 This is a flowchart of a data acquisition and analysis method for safety valve calibration according to the present invention. Detailed Implementation

[0021] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0022] Example 1: As Figure 1 As shown in the embodiment of this application, a data acquisition and analysis system for safety valve calibration is provided, including a dynamic multidimensional data acquisition module, an unsteady-state pressure excitation application module, a multidimensional feature extraction module, a full life cycle baseline management module, a noise separation and identification module, a degradation trend analysis module, and a prediction and early warning module.

[0023] The dynamic multidimensional data acquisition module is used to perform step one, acquire the multidimensional raw signal of the safety valve verification, and apply condition constraints based on the time synchronization mechanism to the multidimensional raw signal; the unsteady pressure excitation application module is used to generate an unsteady pressure excitation signal that matches the dynamic load of the actual working condition of the safety valve. The unsteady pressure excitation application module is used to execute step two, which performs in-depth analysis and feature mining on the multidimensional original signal based on joint time-frequency domain analysis and wavelet multi-scale decomposition, and extracts feature parameters that reflect the dynamic response characteristics of the safety valve. The multidimensional feature extraction module is used to perform step three, which involves in-depth analysis and feature mining of the multidimensional raw signal to extract feature parameters that reflect the dynamic response characteristics of the safety valve. The full lifecycle baseline management module is used to execute step four, which establishes an individualized performance baseline profile based on historical data from each calibration of a single safety valve. The noise separation and identification module is used to perform independent component analysis on the mixed signal, separating the mixed signal into the equipment system noise component and the actual response component of the safety valve; the degradation trend analysis module is used to perform step five, which is based on the full life cycle performance baseline file to perform long-term trend prediction and degradation assessment. The prediction and early warning module is used to execute step six, calculate the remaining service life based on the degradation trend extrapolation model, and generate a judgment result.

[0024] The aforementioned system is deployed at the safety valve calibration station in the industrial field to support intelligent calibration and predictive maintenance decisions for safety valves in high-pressure industrial systems. The system is connected to the main equipment of the calibration bench through a standardized industrial communication interface, receives trigger signals and control commands from the calibration site, and pushes the analysis results and early warning information to the operation and maintenance management system.

[0025] The hardware components of the dynamic multidimensional data acquisition module include a pressure acquisition module, a displacement acquisition module, and a vibration acquisition module; The pressure acquisition module includes at least one set of piezoresistive pressure sensors and signal conditioning circuits. The pressure sensors have a measurement range of 0 to 60 MPa and a measurement accuracy better than 0.05% of the full scale. The signal conditioning circuits include an instrumentation amplifier and a low-pass filter, which are used to amplify and filter the weak voltage signal output by the sensor. The displacement acquisition module includes at least one set of eddy current displacement sensors and a high-speed analog-to-digital converter. The eddy current displacement sensors have a measurement range of 0 to 10 mm and are used to convert the analog output signal of the displacement sensors into a digital signal. The vibration acquisition module includes at least one set of piezoelectric accelerometers and anti-aliasing filter circuits. The frequency response range of the piezoelectric accelerometers is 2 to 10,000 Hz. The cutoff frequency of the anti-aliasing filter circuit is set to half of the sampling frequency to filter out aliasing noise higher than the Nyquist frequency.

[0026] All channels of the above three acquisition modules ensure the time consistency of multi-channel data through a hardware-triggered synchronization mechanism; The data acquisition and control module integrates a multi-channel synchronous sampling controller. After receiving the pressure trigger signal from the calibration bench, the controller starts the sampling operation of all acquisition channels to ensure that the timestamps of the pressure signal, displacement signal and vibration signal are strictly aligned. The sampling frequency and trigger mode can be configured through software.

[0027] The hardware structure of the unsteady-state pressure excitation application module includes a waveform generation module, a power amplification module, and a pressure feedback control module. The waveform generation module is implemented based on a programmable logic device and supports the import and editing of user-defined excitation waveform templates. The waveform generation module internally stores a variety of preset excitation waveform templates, including periodic pressure fluctuation component templates, random pulse component templates, and pressure boost component templates. The output drive current range of the power amplifier module is 0 to 20 amps, which is used to amplify the low-power excitation signal output by the waveform generation module into a power signal that can drive pneumatic or hydraulic actuators. The pressure feedback control module uses a proportional-integral-derivative (PID) control algorithm to achieve rapid and accurate tracking of the target pressure waveform. The pressure feedback control module receives the pressure feedback signal from the pressure acquisition module in real time, calculates the deviation between the target pressure and the actual pressure, and calculates the control quantity according to the PID control algorithm to output to the power amplifier module.

[0028] During the generation of the composite excitation waveform, the waveform generation module reads the weighting coefficients from the preset parameter library based on the rated opening pressure and nominal diameter of the safety valve being tested, and weights and superimposes the periodic pressure fluctuation component, random pulse component and pressure rise component according to the preset weight ratio to generate the composite excitation waveform. The fundamental frequency of the periodic pressure fluctuation component is set to 0.1 to 2 Hz, the peak energy of the random pulse component is concentrated in the range of 100 to 500 milliseconds, and the slope of the pressure rise component varies in the range of 0.01 to 0.5 MPa per second. The weighting coefficients of each component are adaptively adjusted according to the specifications of the safety valve being tested to ensure that the spectral characteristics of the composite excitation waveform are consistent with the dynamic load form that the safety valve is subjected to in the actual high-pressure industrial system.

[0029] The hardware components of the multidimensional feature extraction module include a digital signal preprocessing module, a time-domain analysis module, a frequency-domain analysis module, and a feature fusion module. The digital signal preprocessing module implements sampling rate conversion, digital filtering, and signal denoising functions. This module receives the raw digital signal from the dynamic multidimensional data acquisition module. The resampling algorithm unifies signals with different sampling rates to a common sampling frequency, the finite impulse response digital filter removes power frequency interference and its harmonic components, and the wavelet threshold denoising algorithm further reduces the impact of random noise.

[0030] The time-domain analysis module extracts the statistical and waveform characteristics of pressure and displacement signals; For pressure time-domain data, the time-domain analysis module extracts core parameters such as peak opening pressure, reseating pressure value, pressure settling time, and pressure overshoot. Peak opening pressure refers to the maximum pressure value collected by the pressure sensor at the moment the safety valve opens. The time domain analysis module determines this parameter by performing peak detection algorithm calculations on the pressure time domain signal. The reseating pressure value refers to the pressure value during the pressure stabilization phase after the safety valve reseats. The time-domain analysis module determines this parameter by calculating the statistical mean of the pressure signal during the reseating phase. Pressure build-up time refers to the length of time from when the pressure starts to rise until it reaches a stable value. The time domain analysis module determines this parameter by detecting the time point when the pressure signal crosses a preset threshold and calculating the time difference. Pressure overshoot refers to the percentage by which the pressure peak exceeds the stable value. The time-domain analysis module determines this parameter by calculating the difference between the pressure peak and the stable value and dividing it by the stable value. For displacement time-domain data, the time-domain analysis module extracts dynamic characteristic parameters such as valve opening displacement, maximum movement speed, action response delay time, and reseating closing time. Valve opening displacement refers to the distance the valve disc moves from its initial position to its maximum displacement. The time-domain analysis module determines this parameter by performing extreme value detection on the displacement signal. Maximum speed refers to the maximum speed value during the valve disc's movement. The time domain analysis module determines this parameter by performing differential calculations on the displacement signal and taking the maximum value. Action response delay time refers to the delay time from when the pressure reaches the opening threshold to when the valve disc begins to move. The time domain analysis module determines this parameter by comparing the timestamp difference between the pressure signal and the displacement signal. The reseating closure time refers to the length of time it takes for the valve disc to return from its maximum displacement to its initial position. The time domain analysis module determines this parameter by detecting the time point when the displacement signal returns to the initial position and calculating the time difference.

[0031] The frequency domain analysis module extracts the signal spectral features through fast Fourier transform and wavelet decomposition. The frequency domain analysis module performs fast Fourier transform on the vibration response signal, converts the time domain signal to the frequency domain to obtain the spectral distribution of the vibration signal, and then extracts the vibration dominant frequency as the core frequency feature of the signal. The vibration dominant frequency refers to the frequency position of the frequency component with the largest amplitude in the vibration signal spectrum. The frequency domain analysis module calculates the vibration amplitude as the energy characteristic of the signal. The vibration amplitude refers to the magnitude of the amplitude corresponding to the dominant frequency component in the spectrum of the vibration signal. The frequency domain analysis module also calculates the frequency band energy distribution as the bandwidth characteristic of the signal. The frequency band energy distribution refers to the proportion of energy in each frequency band to the total energy after the vibration signal spectrum is divided into several frequency bands. The frequency domain analysis module uses the db8 wavelet basis function to perform a 6-level multi-scale decomposition of the vibration response signal and extracts the energy characteristics of the decomposition coefficients of each level as a multi-scale representation of the vibration signal. The frequency domain analysis module uses the short-time Fourier transform method to perform joint time-frequency domain analysis. The window function is set to Hanning window, the window length is set to 256 sampling points, and the frequency resolution is set to 10 Hz.

[0032] The feature fusion module correlates and fuses the multi-channel extraction results, outputting a feature vector in a unified format. First, the feature fusion module assembles the pressure feature parameters and displacement feature parameters extracted by the time domain analysis module and the vibration feature parameters extracted by the frequency domain analysis module into a multi-dimensional feature vector according to a unified encoding rule. Then, the feature vector is normalized to map the numerical range of each feature dimension to the interval between 0 and 1. Finally, the normalized feature vector is stored in the feature database and associated with the corresponding verification record.

[0033] The hardware components of the full lifecycle baseline management module include a data storage module, a baseline update module, and a query and retrieval module. The data storage module adopts a distributed time-series database architecture, which supports efficient storage and compressed archiving of massive amounts of verification data. Each verification generates a data packet containing a verification timestamp, device number, verification personnel information, environmental temperature and humidity conditions, original signal file path, feature parameter extraction results, and comprehensive evaluation conclusions. The data packet is verified by hash verification to ensure its integrity and immutability. The baseline update module automatically performs incremental updates to the baseline archive after each verification. The update process includes data verification, feature extraction, and baseline model correction. The query and retrieval module supports historical data queries based on multiple dimensions such as device number, time range, and verification type.

[0034] The full lifecycle baseline management module performs step four of the independent method claim, including: Based on the device number of the safety valve being tested, retrieve all historical test records of the safety valve from the time-series database and construct a sequence of historical feature parameters. The multidimensional feature parameters extracted in the current verification are correlated with the historical feature parameter sequences to calculate the mean, standard deviation and trend of each feature parameter. Compare the current characteristic parameters with the initial factory parameters of the safety valve, and calculate the offset of each characteristic parameter relative to the factory value; The current characteristic parameters are correlated with service information such as the cumulative operating time and cumulative number of opening and closing of the safety valve to establish a mapping relationship between the characteristic parameters and service conditions. Based on the above analysis results, the individualized performance baseline profile of the safety valve is updated. This profile fully records the performance evolution trajectory of the safety valve from its factory to its current state, including the characteristic parameter sequence of each calibration, the trend curve of change, and the comprehensive evaluation conclusion.

[0035] The hardware components of the noise separation and identification module include a signal decomposition module, a noise matching module, and an anomaly detection module. The signal decomposition module implements a blind source separation algorithm based on independent component analysis, which decomposes the mixed signal into statistically independent components. The signal decomposition module constructs the multidimensional original signal into an observation signal matrix, and uses a fast fixed-point algorithm based on maximizing negative entropy to perform independent component analysis and iterative solution, separating the mixed signal into the equipment system noise component and the true response component of the safety valve; The noise matching module identifies the noise components of the equipment system by matching them with templates in the noise feature library. The noise feature library is built based on the statistical average of more than fifty historical normal verification data and stores standard spectrum templates for various equipment system noises. The anomaly detection module compares and analyzes the correlation between changes in noise level and changes in characteristic parameters, and outputs the anomaly type judgment conclusion and confidence index. The cumulative sum control chart method is used to detect changes in noise floor level, and the control limit is set to three standard deviations.

[0036] The noise separation and identification module, when performing independent component analysis on mixed signals, includes: It receives multidimensional raw signals from the dynamic multidimensional data acquisition module and combines pressure signals, displacement signals and vibration signals to construct a multi-channel observation signal matrix. The observed signal matrix is ​​preprocessed, including centering and whitening, the signal mean is reduced to zero and the covariance matrix is ​​diagonalized; Initialize the separation matrix and enter the iteration loop. In each iteration, calculate the update direction of the separation matrix. The update direction is calculated based on the maximization of negative entropy criterion. The approximate calculation of negative entropy adopts the quadratic form of higher-order cumulants. Determine the convergence condition of the separation matrix. Stop iterating when the difference between two consecutive iterations of the separation matrix is ​​less than the convergence accuracy threshold; otherwise, continue iterating and updating. The system outputs a separation matrix and independent component vectors, separates and identifies the noise components in the independent component vectors from the actual response components of the safety valve, and determines whether the noise level of the current verification signal is within the normal range based on the noise reference model. When a feature parameter is detected to deviate from the historical fluctuation range, the system automatically determines whether the deviation is caused by noise interference or substantial performance degradation.

[0037] The hardware components of the degradation trend analysis module include a trend fitting module, a rate calculation module, and a comprehensive evaluation module. The trend fitting module uses multiple time series models for parallel fitting and automatically selects the optimal model based on the fitting residuals. The time series analysis uses an autoregressive integral moving average model, and the model order is automatically determined by the Akaike Information Criterion. The rate calculation module quantifies the degradation rate and acceleration of each characteristic parameter, and the rate of change of the characteristic parameters is calculated using the linear regression slope estimation method; The comprehensive evaluation module calculates the comprehensive degradation index based on a multi-parameter weighted fusion algorithm and generates a trend analysis report. The calculation of the comprehensive degradation index includes weighted terms such as opening pressure offset, valve disc action response delay increment, vibration energy change rate, and seal leakage increment.

[0038] The degradation trend analysis module performs step five of the independent method claim, including: The historical feature parameter time series of the target safety valve is obtained from the full life cycle baseline management module. This time series contains the feature parameter extraction results of each calibration of the safety valve and is arranged in chronological order. An autoregressive integral moving average model is used to fit the trend of the feature parameter time series. The fitting process includes model parameter estimation and model order selection. The model order is automatically determined by the Akaike Information Criterion. The Akaike Information Criterion is calculated by subtracting the ratio of the number of model parameters to the number of samples from the model log-likelihood function value and multiplying it by two. The model order that minimizes the Akaike Information Criterion is selected as the optimal order. The rate of change and acceleration index of the feature parameters are extracted from the fitted model. The rate of change is calculated by the linear regression slope estimation method. Specifically, the time series of the feature parameters are subjected to univariate linear regression analysis, and the slope coefficient of the regression equation is obtained as the rate of change index. Calculate the comprehensive degradation index and determine the weighting coefficients of each characteristic parameter. The weighting coefficients are allocated according to the influence weight of each characteristic parameter on the reliability of the safety valve. The influence weight is determined by statistical analysis of historical failure cases and expert experience. The normalized values ​​of each feature parameter are multiplied by their corresponding weighting coefficients and summed to obtain the value of the comprehensive degradation index. Determine whether the rate of change of the comprehensive degradation index exceeds a preset threshold. When the rate of change exceeds the threshold, a performance warning is automatically triggered. The warning information includes the feature parameter name, degradation rate, warning level, and reference suggestions.

[0039] The hardware components of the prediction and early warning module include a lifespan prediction module, a risk assessment module, and an early warning push module. The lifespan prediction module calculates the remaining lifespan distribution based on a degradation trend extrapolation model, which adopts a combination architecture of a grey prediction model and a neural network prediction model. The risk assessment module determines the warning level based on the relationship between the predicted remaining useful life and the risk threshold. The warning level is divided into three levels: concern, warning, and danger. The early warning push module pushes early warning information and maintenance suggestions to the operation and maintenance management system through multiple communication interfaces.

[0040] The prediction and early warning module performs step six of the independent method claim, including: Obtain the time series of characteristic parameters and the comprehensive degradation index series of the target safety valve from the degradation trend analysis module; A grey prediction model is used to predict the exponential growth trend of the feature parameter sequence. The grey prediction model generates the sequence by accumulating and generating the sequence from the nearest mean, and establishes and solves the grey differential equation to obtain the future predicted value sequence of the feature parameters. A neural network prediction model is used to predict the nonlinear fluctuation characteristics of the feature parameter sequence. The neural network prediction model adopts a three-layer feedforward neural network structure. The number of nodes in the input layer is equal to the number of historical data points of the feature parameter sequence. The number of nodes in the hidden layer is determined by trial and error. The number of nodes in the output layer is equal to the prediction step size. The neural network model is trained using the backpropagation algorithm. The training objective is to minimize the mean square error between the predicted output and the actual value. The outputs of the grey prediction model and the neural network prediction model are weighted and fused. The fusion weights are adaptively adjusted according to the prediction errors of the two models on historical data to obtain the final prediction result sequence. The time point at which the characteristic parameters reach the failure threshold is calculated based on the predicted result sequence. The failure threshold is determined according to the technical specifications and industry standards of the safety valve. The calculation method is to compare the predicted value sequence of the characteristic parameters with the failure threshold, find the predicted time point when the predicted value first exceeds the failure threshold, and take this time point as the failure prediction time of the characteristic parameter. The minimum value of the failure prediction time of each characteristic parameter is taken as the remaining service life prediction value of the safety valve. The prediction result includes point estimates and confidence intervals. The confidence intervals are calculated using the Monte Carlo simulation method. The warning level is determined based on the relationship between the predicted remaining useful life and the risk threshold. A remaining useful life greater than 90 days is classified as a concern level, 30 to 90 days as a warning level, and less than 30 days as a danger level. The warning information is pushed to the operation and maintenance management system through the Ethernet communication interface. The warning content includes the equipment identification, warning level, prediction conclusion, and reference maintenance suggestions.

[0041] like Figure 2 As shown, the data acquisition and analysis method for safety valve calibration according to the present invention includes step one: acquiring the multidimensional raw signal of the safety valve calibration. The dynamic multidimensional data acquisition module responds to the verification start trigger signal, and the specific execution process of this step is as follows: The dynamic multidimensional data acquisition module first activates the pressure acquisition module, displacement acquisition module and vibration acquisition module through control commands, so that they enter the data acquisition ready state. The piezoresistive pressure sensor of the pressure acquisition module acquires the instantaneous value of the inlet pressure in front of the valve in real time and outputs the corresponding voltage signal. After being amplified and filtered by the signal conditioning circuit, it is transmitted to the high-speed analog-to-digital converter and converted into a digital pressure signal. The sampling frequency is 2,500 sampling points per second. The eddy current displacement sensor of the displacement acquisition module measures the displacement change of the valve disc relative to the initial position in real time and outputs the corresponding voltage signal. After being processed by the signal conditioning circuit, it is transmitted to the high-speed analog-to-digital converter and converted into a digital displacement signal. The sampling frequency is 2,500 sampling points per second. The piezoelectric accelerometer of the vibration acquisition module collects the vibration response signal of the safety valve housing in real time during the operation process, outputs the corresponding charge signal, and after being processed by the charge amplifier and anti-aliasing filter circuit, it is transmitted to the high-speed analog-to-digital converter to be converted into a digital vibration signal. The sampling frequency is 5,000 sampling points per second. After receiving the pressure trigger signal, the data acquisition and control module simultaneously starts the sampling operation of the three acquisition modules with microsecond-level synchronization accuracy to ensure that the timestamps of the multi-channel digital signals are strictly aligned.

[0042] Step 2: Apply time synchronization-based constraints to the multidimensional original signal. After acquiring the data, the dynamic multidimensional data acquisition module continues to execute step two; The specific execution process for this step is as follows: The data acquisition and control module adds a unique timestamp to each sampling point. The timestamp is generated based on a high-precision real-time clock chip to ensure the time alignment accuracy of all channel sampling data. The data acquisition and control module performs synchronous correction processing on the multi-channel raw signals, specifically including sampling rate unification and phase alignment. Sampling rate unification uses a resampling algorithm to unify signals with different raw sampling rates to a common sampling frequency, and phase alignment uses interpolation to adjust the signals of each channel to the same phase reference. The data acquisition and control module performs integrity verification on the synchronized and corrected multidimensional raw signals. It uses check codes to ensure that there are no packet losses or out-of-order transmissions during the data transmission process. All multidimensional raw signals that have undergone time synchronization constraint processing are completely archived and stored in the local database, providing a time-aligned data foundation for subsequent feature extraction and analysis.

[0043] Step 3: Perform in-depth analysis and feature mining on the multidimensional raw signal to extract feature parameters reflecting the dynamic response characteristics of the safety valve. After receiving the time-synchronized original multidimensional signal, the multidimensional feature extraction module executes step three; the specific execution process of this step is as follows: The digital signal preprocessing module performs preprocessing operations on the multidimensional raw signal. It filters out power frequency interference and its harmonic components in each channel signal through a finite impulse response low-pass filter. The filter's passband cutoff frequency is set to 400 Hz, the stopband cutoff frequency is set to 500 Hz, and the minimum stopband attenuation is set to 60 dB. The influence of random noise is further reduced by wavelet thresholding denoising algorithm. The wavelet basis function is db8 wavelet, the number of decomposition layers is set to 5, and the threshold is selected by soft thresholding method. The time-domain analysis module performs feature extraction on the preprocessed pressure and displacement time-domain data. When extracting the opening pressure peak, a peak detection algorithm is used to search for the maximum value of the signal during the safety valve opening period and take the corresponding pressure value as the opening pressure peak. When extracting the reseating pressure value, an average value calculation method is used to calculate the average value of the pressure signal during the reseating stabilization period. When extracting the pressure settling time, a threshold crossing detection method is used to calculate the time difference between the pressure signal rising from the initial value to the stable value. When extracting the valve disc opening displacement, an extreme value detection algorithm is used to search for the maximum value in the displacement signal and calculate the difference with the initial value. When extracting the action response delay time, a cross-correlation analysis method is used to calculate the time lag between the pressure signal and the displacement signal. When extracting the reseating closing time, a threshold crossing detection method is used to calculate the time difference between the displacement signal returning from the maximum value to the initial value. The frequency domain analysis module extracts frequency domain features from the preprocessed vibration response data and converts the vibration time domain signal to the frequency domain through fast Fourier transform to obtain a spectrum sequence. Search for the frequency point with the largest amplitude in the spectrum sequence and take the corresponding frequency value as the vibration main frequency. When extracting the vibration amplitude, calculate the amplitude of the frequency point corresponding to the vibration main frequency. When extracting the frequency band energy distribution, divide the vibration spectrum into several frequency bands and calculate the energy proportion in each frequency band. When extracting the vibration energy change rate, compare the current verified vibration energy with the historical verified vibration energy to calculate the percentage change. The feature fusion module assembles the time-domain analysis results and frequency-domain analysis results into a multi-dimensional feature vector according to a unified coding rule. The feature vector contains core parameters such as peak opening pressure, reseating pressure value, pressure settling time, valve opening displacement, action response delay time, vibration dominant frequency, and frequency band energy distribution. After normalization, the feature vector is stored in the feature database and associated with the corresponding verification record.

[0044] Step 4: Establish an individualized performance baseline profile based on historical data from each calibration of a single safety valve. Perform correlation analysis between the extracted feature parameters and the initial factory parameters and historical calibration records to construct a performance evolution trajectory reflecting the safety valve's performance from factory to current state. The full lifecycle baseline management module executes step four; the specific execution process of this step is as follows: Based on the equipment number of the safety valve being calibrated, retrieve the factory initial parameter file of the safety valve from the time series database. The factory initial parameters include the rated opening pressure, rated reseating pressure, rated discharge capacity, and the benchmark characteristic parameter values ​​for factory calibration. Retrieve all historical verification records of the safety valve from the feature database, construct a time series of historical feature parameters, and the historical verification records include the timestamps of each verification, the results of feature parameter extraction, and the comprehensive evaluation conclusions. The multidimensional feature parameters extracted during the current verification are compared with the initial parameters at the factory, and the absolute and percentage offsets of each feature parameter relative to the factory value are calculated. The offset calculation results reflect the degree of deviation of the current performance of the safety valve from the factory state. The current feature parameters are correlated with the historical feature parameter time series. The mean, standard deviation, extreme values ​​and long-term trend of each feature parameter in the historical time series are calculated. The trend analysis uses linear regression to fit the functional relationship between the feature parameters and the cumulative running time. The current characteristic parameters are correlated with the service information of the safety valve, such as the cumulative operating time, cumulative number of opening and closing times, and cumulative number of maintenance times. A mapping relationship table between the characteristic parameters and the service conditions is established. The mapping relationship table records the change pattern of each characteristic parameter with the cumulative operating time and number of opening and closing times. Based on the above analysis results, the individualized performance baseline file of the safety valve is updated. The data structure of the performance baseline file includes the equipment basic information area, the factory parameter area, the calibration record index area, the historical feature sequence area, the change trend model area, and the comprehensive evaluation area. The calibration record index area maintains a pointer sequence pointing to historical calibration records. The historical feature sequence area stores a two-dimensional matrix composed of feature parameter vectors from each calibration in chronological order. The change trend model area stores the time series fitting model parameters of each feature parameter. The performance baseline profile ensures data integrity through hash verification. Each time it is updated, the hash value of the profile data packet is recalculated and stored for subsequent data integrity verification.

[0045] Step 5: Based on the full lifecycle performance baseline file, perform long-term trend prediction and degradation assessment, calculate the comprehensive degradation index, and trigger an alert when the rate of change of the comprehensive degradation index exceeds a preset threshold. The degradation trend analysis module executes step five; the specific execution process of this step is as follows: The individualized performance baseline profile of the target safety valve is obtained from the full life cycle baseline management module, and the time series data of the feature parameters in the historical feature sequence area of ​​the profile are extracted. The trend fitting module uses an autoregressive integral moving average model to fit the trend of the feature parameter time series, performs difference operations on the time series to make it stable, estimates the autoregressive coefficients and moving average coefficients of the model through the maximum likelihood estimation method, and automatically determines the specific value of the model order through the Akaike Information Criterion. The trend fitting module uses an exponential smoothing model and a grey prediction model for parallel fitting. The fitting accuracy of each model is evaluated by calculating the mean square error of the fitting residuals. The optimal model is automatically selected as the final trend fitting model based on the principle of minimizing the fitting residuals. The rate calculation module extracts the rate of change and acceleration of each feature parameter from the optimal trend fitting model. The rate of change is calculated by the slope coefficient of the fitting curve, which represents the value of the feature parameter change with unit time. The acceleration is calculated by the second derivative of the fitting curve, which represents the trend of the rate of change with time. The rate calculation module compares and analyzes the current value, rate of change, and acceleration of each characteristic parameter with the corresponding failure threshold to identify the valve disc movement rate decay trend, action response delay increase trend, and sealing performance decline trend. When the change trend of a certain characteristic parameter points to the failure threshold, it is marked as a potential degradation risk item. The comprehensive evaluation module calculates the comprehensive degradation index. First, it determines the weighting coefficient of each characteristic parameter in the calculation of the comprehensive degradation index. The weighting coefficient is allocated according to the degree of influence of each characteristic parameter on the reliability of the safety valve function. The degree of influence is determined by the failure mode analysis of historical failure cases. The degradation of the valve disc opening displacement directly affects the pressure relief function and is therefore given a high weight. The increase in action response delay will affect the timeliness of action and is therefore given a medium weight. Abnormal changes in vibration characteristics reflect the loosening of the internal structure and are therefore given a certain weight. The weighting coefficient for the valve opening displacement is set to a range of 0.45 to 0.65, with a specific example of 0.55. This is based on the fact that in historical failure cases, the proportion of pressure relief failures caused by insufficient valve displacement is 52%-68%, and when the displacement degrades to below 80% of the baseline average, the effective pressure relief area will decrease by more than 20%, which will lead to an exponential increase in the risk of system overpressure. The weighting factor for action response delay time is set to the range of 0.20 to 0.30, with a specific value example of 0.25. This is based on the fact that in historical failure mode analysis, the proportion of overpressure accidents caused by increased response delay is about 15%-25%, and when the response delay increases from the baseline average of 50 milliseconds to 80 milliseconds, it will shorten the buffer time for system pressure rise, which is a medium-risk factor. The weighting coefficient for the sealing leakage rate is set to a range of 0.12 to 0.20, with a specific example of 0.15. This is based on the fact that in historical failure statistics, accidents caused by sealing deterioration account for approximately 10%-18%, and that an increased leakage rate accelerates the erosion of the sealing surface, which is an early warning indicator. The weighting coefficient for the change in vibration characteristic frequency amplitude is set to a range of 0.05 to 0.12, with a specific example of 0.05. This is based on the fact that in historical failure mode analysis, vibration anomalies appear as auxiliary signs in about 5%-12% of failure cases, and vibration characteristics are greatly affected by environmental noise. Anomalies often lag behind displacement and leakage parameters and are therefore considered auxiliary weights. The sum of the above four weighting coefficients is strictly equal to 1, and the specific value of each coefficient is fine-tuned during initialization based on the safety valve model and service conditions using the analytic hierarchy process. The judgment matrix of the analytic hierarchy process is determined based on the consistency test of the scores of at least 5 senior fault diagnosis experts, and the consistency ratio must be less than 0.1. In terms of stationarity testing, the enhanced Dickey-Fowler test has been newly adopted. The series is judged to be stationary when the test statistic is less than the critical value at a significance level of 0.05. Regarding model order, the range of values ​​for autoregressive order and moving average order is set to 0 to 5. The combination of orders with the lowest Akaike Information Criterion value is the final model order. The calculation method for the normalized degradation degree value is refined to divide the absolute value of the difference between the current feature parameter value and the baseline mean by the difference between the failure threshold and the baseline mean. If the current value is between the baseline mean and the failure threshold, the normalized value is between 0 and 1. If the current value exceeds the failure threshold, the normalized value is 1. If the current value is better than the baseline mean, the normalized value is 0. The range of the comprehensive degradation index is 0 to 1. The larger the value, the more severe the degradation. The calculation of the rate of change has been updated to include a new feature: if the comprehensive degradation index of the previous verification period is zero, the average rate of change of the three consecutive periods will be used instead. Regarding the rate of change threshold, a new feature has been added: based on the fluctuation range of historical normal operation data, the daily rate of change should not exceed 15% or the weekly rate of change should not exceed 40%. In a specific embodiment, the daily rate of change threshold is set to 12%. Regarding the warning levels, a new system has been added that classifies the comprehensive degradation index into three levels based on the combination of the absolute value and the rate of change. Specifically, when the comprehensive degradation index reaches 0.7 or above or the rate of change reaches 30% or above, a Level 1 warning is triggered for emergency maintenance; when the comprehensive degradation index is between 0.5 and 0.7 and the rate of change reaches 15% or above, a Level 2 warning is triggered for planned maintenance; and when the comprehensive degradation index is between 0.3 and 0.5 and the rate of change reaches 12% or above, a Level 3 warning is triggered for status monitoring. The specific calculation process of the comprehensive degradation index is to multiply the normalized degradation degree value of each characteristic parameter by the corresponding weighting coefficient and then sum them. The normalized degradation degree value is calculated by the ratio of the deviation of the current characteristic parameter value from the baseline mean to the distance of the failure threshold. The comprehensive evaluation module calculates the rate of change of the comprehensive degradation index. The rate of change is obtained by dividing the difference between the current comprehensive degradation index and the comprehensive degradation index of the previous verification period by the comprehensive degradation index of the previous verification period. When the rate of change exceeds the preset threshold, a performance warning is automatically triggered. The warning information includes the warning level, the name of the characteristic parameter that triggered the warning, the quantified value of the degradation rate, and reference maintenance suggestions.

[0046] Step Six: Calculate the remaining useful life based on the degradation trend extrapolation model, and generate a judgment result directly related to the qualitative condition of the industrial site. The prediction and early warning module executes step six; the specific execution process of this step is as follows: The module obtains the time series of characteristic parameters, the comprehensive degradation index series, and the degradation trend fitting model of each characteristic parameter of the target safety valve from the degradation trend analysis module. The lifespan prediction module calculates the remaining lifespan distribution based on a degradation trend extrapolation model, which adopts a combination architecture of a grey prediction model and a neural network prediction model. The construction process of the grey prediction model includes accumulating the time series of feature parameters to obtain the accumulated sequence, establishing a first-order linear grey differential equation based on the exponential growth characteristic of the accumulated sequence, estimating the coefficients of the grey differential equation using the least squares method, solving the grey differential equation to obtain the time response function, and predicting the future evolution trend of the feature parameters accordingly. The process of constructing a neural network prediction model is as follows: First, the feature parameter time series is constructed into input-output sample pairs for the neural network. The input is a continuous historical data window, and the output is the predicted value at the next time step. The activation function of the hidden layer is the hyperbolic tangent function, and the activation function of the output layer is a linear function. The neural network parameters are trained and optimized through the error backpropagation algorithm until convergence. The lifespan prediction module weights and fuses the output of the grey prediction model with the output of the neural network prediction model to obtain a combined prediction result. The process of determining the fusion weight includes using the two models to backtrack and predict historical data, calculating the root mean square value of each prediction error, and determining the fusion weight based on the reciprocal of the root mean square value of the error. The life prediction module calculates the failure prediction time of each characteristic parameter based on the combined prediction results. The specific process is to recursively calculate the prediction sequence from the current moment, find the time point when the predicted value of each characteristic parameter first exceeds the corresponding failure threshold, and use the number of days from the current moment as the remaining life prediction value of the characteristic parameter. The life prediction module takes the minimum value of the remaining service life prediction of each characteristic parameter as the final prediction result of the remaining service life of the safety valve. At the same time, the Monte Carlo simulation method is used to calculate the confidence interval of the prediction result. The specific process of Monte Carlo simulation is to add a random disturbance that follows a normal distribution to the model parameters and repeat the prediction calculation one thousand times. The 95% confidence interval is calculated based on the distribution of the prediction results. The risk assessment module determines the warning level based on the relationship between the predicted remaining useful life and the risk threshold. The risk threshold is set in three levels according to industry safety standards and the importance of the equipment: If the remaining service life is predicted to be greater than 90 days, it is classified as a concern level, and the warning content is to recommend that a verification and inspection be arranged at the next planned shutdown. If the remaining service life is between 30 and 90 days, it is classified as a warning level, and the warning content is to recommend that on-site inspection and maintenance preparation be arranged within 30 days. If the remaining service life is less than 30 days, it is classified as a danger level, and the warning content is to recommend that an immediate shutdown for maintenance be arranged and spare parts be prepared for replacement. The early warning push module encapsulates early warning information into a standardized message format and pushes it to the operation and maintenance management system via an Ethernet communication interface. The early warning message includes a device identification field that records the device number and installation location of the safety valve, an early warning level field that records the determined early warning level, a prediction conclusion field that records the point estimate and confidence interval of the remaining service life, a reference basis field that records the name of the characteristic parameter that triggered the early warning and the quantification value of the degree of degradation, and a maintenance suggestion field that records the maintenance action guidance corresponding to the early warning level.

[0047] Example 2: When the specifications and rated operating parameters of the safety valve being tested received by the system differ from the preset standard parameters, the waveform generation module of the unsteady pressure excitation application module automatically enters the adaptive adjustment mode. The waveform generation module retrieves the rated opening pressure, nominal diameter, and applicable medium type of the safety valve being tested from the equipment parameter database, and determines the initial weighting coefficient configuration of the composite excitation waveform based on these parameters.

[0048] After generating the composite excitation waveform, the waveform generation module performs parameter fine-tuning through actual application verification by the pressure feedback control module; The initially set composite excitation waveform is applied to the verification pipeline and the actual pressure response curve is collected. The actual pressure response curve is compared and analyzed with the target excitation waveform. When there are deviations between the two in key indicators such as peak time, rise rate and oscillation characteristics, the waveform generation module iteratively optimizes and adjusts the weighting coefficient according to the deviation until the actual response curve is close to the target waveform. The adaptively adjusted excitation waveform parameters are recorded and associated with the calibration configuration file of the safety valve of this specification and model, so that they can be directly called when calibrating the same model of safety valve in the future. This modified solution enables the system to adapt to the calibration requirements of safety valves of different specifications and models, ensuring that the excitation waveform matches the working characteristics of the object being calibrated, thereby improving the accuracy and reliability of the calibration results.

[0049] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A data acquisition and analysis method for safety valve calibration, characterized in that, Includes the following steps: Step 1: Obtain the multidimensional raw signals for safety valve calibration; Step 2: Apply time synchronization-based constraints to the multidimensional original signal; Step 3: Perform in-depth analysis and feature mining on the multidimensional raw signals to extract feature parameters that reflect the dynamic response characteristics of the safety valve; Step 4: Based on the historical data of each safety valve's calibration, establish an individualized performance baseline profile, and perform correlation analysis between the extracted feature parameters and the initial factory parameters and historical calibration records to construct a performance evolution trajectory that reflects the safety valve from its factory state to its current state. Step 5: Based on the full lifecycle performance baseline file, perform long-term trend prediction and degradation assessment, calculate the comprehensive degradation index, and trigger an early warning when the rate of change of the comprehensive degradation index exceeds the preset threshold; Step 6: Calculate the remaining useful life based on the degradation trend extrapolation model and generate the judgment result.

2. The data acquisition and analysis method for safety valve calibration according to claim 1, characterized in that, In step three, in-depth analysis and feature mining are performed on the multidimensional original signal. The peak value of opening pressure, the value of reseating pressure, the pressure establishment time, the valve opening displacement, the action response delay time, the vibration main frequency and the frequency band energy distribution parameters are extracted through joint time-frequency domain analysis and wavelet multi-scale decomposition.

3. The data acquisition and analysis method for safety valve calibration according to claim 2, characterized in that, In step five, the characteristic parameter sequence is fitted with a time series analysis algorithm to extract the rate of change and acceleration index of the performance parameters throughout the entire life cycle, identify the decay trend of valve disc movement rate, the increase trend of action response delay and the decline trend of sealing performance, and calculate the comprehensive degradation index using a multi-parameter fusion evaluation mechanism. A performance warning is triggered when the rate of change of the comprehensive degradation index exceeds a preset threshold.

4. A data acquisition and analysis system for safety valve calibration, applicable to the data acquisition and analysis method for safety valve calibration as described in any one of claims 1-3, characterized in that, include: The dynamic multidimensional data acquisition module acquires the multidimensional raw signals of the safety valve verification and applies conditional constraints based on a time synchronization mechanism to the multidimensional raw signals. The unsteady-state pressure excitation application module performs in-depth analysis and feature mining of multidimensional original signals based on joint time-frequency domain analysis and wavelet multi-scale decomposition, and extracts feature parameters that reflect the dynamic response characteristics of the safety valve. The multi-dimensional feature extraction module establishes an individualized performance baseline profile based on historical data from each calibration of a single safety valve, and performs correlation analysis between feature parameters and initial factory parameters and historical calibration records to construct a performance evolution trajectory. The full lifecycle baseline management module performs multi-parameter fusion evaluation of performance evolution trajectory and calculates comprehensive degradation index; When the rate of change of the comprehensive degradation index exceeds the preset threshold, an early warning is triggered. The remaining service life is calculated based on the degradation trend extrapolation model, and the judgment result is output, which is linked to the qualitative status of the industrial site.

5. The data acquisition and analysis system for safety valve calibration according to claim 4, characterized in that, The process of acquiring the multidimensional raw signal for safety valve verification also includes: Pressure acquisition module, which acquires the instantaneous value of the inlet pressure before the valve; The displacement acquisition module tracks the displacement trajectory of the valve disc during the opening and closing process; The vibration acquisition module captures the vibration response characteristics of the shell, and all acquisition channels ensure the time consistency of multi-channel data through a hardware-triggered synchronization mechanism.

6. The data acquisition and analysis system for safety valve calibration according to claim 4, characterized in that, The multidimensional raw signals acquired by the dynamic multidimensional data acquisition module are separated into noise components and the true response components of the safety valve through independent component analysis, and a noise benchmark reference model is established.

7. The data acquisition and analysis system for safety valve calibration according to claim 4, characterized in that, In the process of establishing individualized performance baseline profiles, the multidimensional feature parameters extracted from each verification are correlated with the initial factory parameters, cumulative operating time, and historical verification records of the safety valve; Construct an inverse performance evolution trajectory and identify the initial differences between individual products of the same model by establishing a mapping relationship between characteristic parameters and service conditions.

8. A data acquisition and analysis system for safety valve calibration according to claim 4, characterized in that, It also includes a noise separation and identification module, which performs independent component analysis on the mixed signal after acquiring the multidimensional raw signal of the safety valve verification; When a feature parameter is detected to deviate from its historical fluctuation range, an anomaly type determination conclusion is output.

9. A data acquisition and analysis system for safety valve calibration according to claim 4, characterized in that, It also includes a degradation trend analysis module, which uses multiple time series models for parallel fitting in the process of multi-parameter fusion evaluation of performance evolution trajectory; The optimal model is automatically selected based on the fitting residuals, and the degradation rate and acceleration of each feature parameter are quantified. Based on a multi-parameter weighted fusion algorithm, a comprehensive degradation index is calculated and a trend analysis report is generated.

10. A data acquisition and analysis system for safety valve calibration according to claim 4, characterized in that, It also includes a prediction and early warning module. In the process of calculating the remaining useful life based on the degradation trend extrapolation model, the degradation trend extrapolation model adopts a combination architecture of gray prediction model and neural network prediction model to calculate the distribution of remaining useful life. The prediction results include point estimates and confidence intervals.

Images